The Dialect of Elis and Its Position Within the Greek Dialectological System MA-Thesis for the Master Classics and Ancient Civilisations

Home , Ancient Greek dialects, Doris (Greece), Elis (city), Optative (Ancient Greek)

The dialect of Elis and its position within the Greek dialectological system MA-thesis for the Master Classics and Ancient Civilisations

Le site antique d’Olympie, illustration taken from Minon 2007 : 559 by M.J. van der Velden BA supervisors: dr. L. van Beek dr. A. Rademaker

2015-17 Universiteit Leiden

Table of contents i. Acknowledgements ii. List of abbreviations

0. Introduction

1. The dialect features of Elean 1.1 West Greek features 1.1.1 West Greek phonological features 1.1.2 West Greek morphological features 1.1.3 Conclusion

1.2 Northwest Greek features 1.2.1 Northwest Greek phonological features 1.2.2 Northwest Greek morphological features 1.2.3 Conclusion

1.3 Features in common with various other dialects 1.3.1 Phonological features in common with various other dialects 1.3.2 Morphological features in common with various other dialects 1.3.3 Conclusion

1.4 Specifically Elean features 1.4.1 Specifically Elean phonological features 1.4.2 Specifically Elean morphological features 1.4.3 Conclusion

1.5 General conclusion

2. Evaluation 2.1 The consonant stem accusative plural in -ες 2.2 The consonant stem dative plural endings -οις and -εσσι 2.3 The middle participle in /-ēmenos/ 2.4 The development *ē > ǟ 2.5 The development *ӗ > α 2.6 The development *i > ε

3. Conclusion

4. Bibliography

Acknowledgements

First of all, I would like to express my deepest gratitude towards Lucien van Beek for supervising my work, without whose help, comments and – at times necessary – incitations this study would not have reached its current shape, as well as towards Adriaan Rademaker for carefully reading my work and sharing his remarks.

Secondly, I would like to thank all the teachers from Classics staff for teaching me and raising my Greek and Latin language proficiency to the level it currently has, and in particular Joëlle Koning, Adriaan Rademaker, Marjolein van Raalte, and Ineke Sluiter for helping me paving the path into the Master’s Programme of Classics and Ancient Civilisations.

Finally, my friends’ and family’s emotional support must not be underestimated, carrying me through the process of writing this work at times things were going less fluently as they should. Also to them I wish to express my utmost gratitude.

Leiden, May 2017

List of abbreviations acc. - accusative aor. - aorist art. - article Att. - Attic comp. - comparative conj. - conjunction Cr. - Cretan dat. - dative deict. - deictic du. - dual fem. - feminine fut. - future gen. - genitive imp. - imperative inf. - infinitive Locr. Ep. - Epizephyrian Locrian masc. - masculine mid. - middle n. - noun neg. - negative neut. - neuter nom. - nominative opt. - optative p. - person p. pron. - personal pronoun par. - paragraph part. - participle Pgr. - Proto-Greek PIE - Proto-Indo-European pl. - plural prep. - preposition pres. - present ptc. - participle sg. - singular Skt. - Sanskrit subj. - subjunctive superl. - superlative

0. Introduction

The Greek dialect of Olympia and the surrounding region of Elis was spoken in the northwest of the Peloponnese and has been transmitted to us by thirty-four larger inscriptions. The majority of these date from the sixth and fifth century B.C. A smaller part of eleven inscriptions from the next two centuries – in which an increase of similarity with the koinè language is discernible – is also attested.

The first inscription written in the Elean dialect – a bronze tablet containing the conditions of a treaty between the Olympians and the neighbouring town of Eua – was discovered by William Gell in 1813. Numerous other inscriptions from Elis have been found since then, so that the present number of the Elean inscriptions handed down to us is over thirty, which is a fairly rich amount in comparison to the number of inscriptions from other Greek cities at such an early stage. Apart from these larger inscriptions, which mainly contain (religious) law decrees and treaties, smaller texts written in the Elean dialect, such as dedications and signatures, have also been found.1

Some preliminary methodological remarks are in order. In this study, I will make an attempt to classify the Elean dialect genetically. That is, I will investigate from which dialect group of Greek it has most probably descended. This is not necessarily the group with which it shares the greatest number of features synchronically; as we will see, some of these dialect features were probably secondarily taken over from other dialects. The number of features shared with other dialects is only of interest for a synchronic dialect classification.

The ancient Greek dialects have been classified into East- and West Greek. The first group comprises Attic-Ionic, Arcado-Cyprian, Aeolic, and Mycenaean. West Greek consists of Doric and Northwest Greek.2 To the Doric subgroup belong for instance Cretan, Laconian, Arcadian, and Corinthian. Northwest Greek is made up by Phocian, Locrian, Aetolian, Acarnanian, and Epirotic.3

As far as East Greek is concerned, however, Risch (1955) has questioned the grouping of Aeolic under this dialect group. On the basis of relative chronology, he argues that the features which Aeolic shares with East Greek – which are found mostly in Lesbian – are relatively young. According to Risch, these features have found their way into Lesbian by language contact with Ionic, which was geographically close. Hence, parts of Boeotian and Thessalian are the more conservative Aeolic dialects, making the grouping of Aeolic alongside with Doric and Northwest Greek more plausible, Risch argues. This leads him to a regrouping of the ancient Greek dialects into North and South Greek rather than into an East Greek and a West Greek group. In his classification, North Greek consists of Doric, Northwest Greek, and Aeolic, while South Greek comprises Ionic-Attic, Arcado-Cyprian, and Mycenaean. In this thesis, the term West Greek will be used in order to refer to Doric and Northwest Greek as a whole.

1 Minon 2007 : 1-3.

2 Buck 1955 : 7.

3 Thumb-Kieckers 1932 : 251.

Scholars agree that the dialect of Greek which is represented by the inscriptions from Elis is a North Greek one, but controversy still exists over the more precise classification of Elean within this dialect group. This disagreement stems from the fact that Elean has shared features not only with Doric, but also with Northwest Greek and Aeolic. Apart from that, Elean also shows some remarkable features of its own.4

Buck has classified Elean under the Northwest Greek subgroup.5 Méndez Dosuna, however, places Elean outside of Northwest Greek and gives preference to an origin within Doric.6 Minon follows Méndez Dosuna’s conclusion, but nevertheless points to some striking isoglosses with Locrian.7 In Bartonĕk’s view, Elean is a separate descendant from Proto-West Greek.8 In view of this lack of consensus, there is every reason to reconsider the historical classification of the Elean dialect.

For a genetic classification of Elean, it is of crucial importance to determine with which dialects it shares ancient features. In order to do this, we have to investigate the age of the dialect features of Elean. This will be done on the basis of relative chronology. That is, it will be determined which sequence of sound changes is the more probable one, and, on this basis, which dialect features are the oldest.

In the first chapter, a wide range of phonological and morphological features of Elean will be presented and briefly discussed. In chapter 2, a selection of these features is more thoroughly discussed and evaluated. On this basis, I present my preliminary conclusion concerning the dialect position of Elean in chapter 3.

Special attention will be paid to shared innovations, which provide a stronger indication of dialectal relatedness than shared archaisms do.

However, as García Ramón (2009) argues, the importance of shared archaisms should not be underestimated. In a thorough discussion of Parker’s (2008) arguments against the unity of Aeolic, García Ramón warns against too skeptic an approach when using innovations and archaisms to determine dialectal relatedness. He uses Arcado-Cyprian as a case study. Parker has argued that Arcadian and Cyprian do not form one branch within South Greek because of a lack of ‘demonstrable common innovations’ between the two dialects.9 However, as García Ramón points out, Parker ignores the fact that the retention of a certain feature may indicate dialectal relatedness as well, especially in cases where other dialects have undergone an innovation. For example, the interrelated dialects of Boeotian and Thessalian both use the archaistic patronymic adjective, whereas the dialects

4 One of the most striking of these is word-final rhotacism, i.e. the representation of *-s by <ρ>, found sporadically in the ancient inscriptions but consistently in the more recent ones. The only other ancient Greek dialects which show rhotacism are Eretrian and late-Laconian. Rhotacism lives forth in the Modern Greek Tsakonian dialect.

5 Buck 1955 : 11.

6 Méndez Dosuna 1980 : 181, 200.

7 Minon 2007 : 629.

8 Bartonĕk 1972 : 59-65.

9 García Ramón 2009 : 227.

that are situated in between them do not.10 Hence, we should not dismiss shared archaisms between Elean and other dialects too easily: they may provide some indication of relatedness. In what follows, I will therefore, in addition to the innovations, also discuss the retained archaisms of Elean.

The overview of dialect characteristics presented in the first chapter is taken from Buck (1955). For the quotation of material from the inscriptions, which have been published by Dittenberger (Die Inschriften von Olympia) in 1896 and more recently by Minon (2007), I will primarily make use of the older inscriptions from the sixth and fifth century B.C. The younger inscriptions will be used in those cases where no material concerning a certain feature is found in the older ones or in cases where conclusions that may be drawn from the younger inscriptions are in contradiction to those drawn from the older ones. In quotations from Elean, I primarily use the diacritic signs as they are applied by Minon. In quotations from articles written by other authors, I chose to leave the diacritics the same as they are displayed by the respective authors.

10 García Ramón 2009 : 222.

1. The dialect features of Elean

In this chapter I present a fourfold overview of the linguistic features that define the Elean dialect. In the first paragraph, the features that Elean shares with its West Greek counterparts will be discussed. Next come the ones that Elean has in common with Northwest Greek. In the third paragraph, the features which Elean has in common with various other, not specifically Northwest Greek dialects are dealt with. In the fourth, those that are specifically Elean will be presented.

The lists of selected dialect features is based on Buck’s discussion of Elean dialect features.11 For their discussion, I mainly used Minon (2007), Méndez Dosuna (1985) and Buck (1955). In those cases where other sources than these three handbooks are used, this will be indicated in the footnotes. The main aim of the discussion of the features is to evaluate whether a particular feature can be relevant for the goal of this thesis, i.e. the genetic classification of the Elean dialect.

The discussion of the features will be illustrated by examples quoted from the Elean inscriptions. Each quote is followed by the number of the particular inscription and the line in which the present form is found. For the sake of clarity, a classical Greek cognate of the quoted form will be given as well.

Note that in some cases, it is probable that Elean did have a certain feature, although it is not attested in the texts handed down to us. An example of this is the 1st person plural form in -μες, which is characteristically North Greek, rather than the -μεν ending, which is typical for South Greek. Another example is the North Greek form δήλομαι (δείλομαι) as an equivalent to South Greek βούλομαι, with e-grade rather than o-grade and thence development of the Proto-Greek labiovelar *gʷ- to a dental δ- instead of a labial β- in South Greek.

1.1 West Greek features

In this paragraph, I will present the phonological and morphological features that Elean has in common with the other West Greek dialects. Since there is consensus that Elean is a West Greek dialect, these features are not so much of interest for our aim to arrive at a more precise classification of the dialect within West Greek. The features listed in this section merely confirm the character of Elean as a West Greek dialect.

1.1.1 West Greek phonological features

Three phonological features that confirm the West Greek identity of Elean are:

1) the retention of *-ti- as -τι- 2) the a vocalism in the outcome of the sequence *-km̥ t- in numerals such as Ϝίκατι ‘twenty’, and 3) the a vocalism in ἱαρός ‘sacred’.

11 Buck 1955 : 154-60.

The Proto-Greek sequence *-ti- assibilated to -σι- in South Greek.12 This is seen in for instance, the 3rd pl. pres. μετέχουσι. In West Greek (and therefore also in Elean) however, this development did not take place, as shown by the cognate form μετέχοντι [34.18]13. The assibilation of *-ti- must be dated in the second millennium B.C., for, as appears from the above quotations, it preceded the loss of -ν- before -σ- and it has been attested in Mycenean.14 For this reason, it is one of the most important isoglosses separating West Greek from South Greek.

Another phonological feature that separates the West Greek dialects from South Greek is the a vocalism in the suffixes of certain numerals. West Greek shows a vocalism in the suffix of Ϝίκατι ‘twenty’ and in that of the higher ordinals in -κάτιοι (with the West Greek retention of -τι- just discussed). An example of this in Elean is seen in the acc. pl. πεντακατίας ‘five hundred’ [9.2-3], corresponding to South Greek πεντακοσίας. As becomes clear from the latter form, o vocalism in the suffixes of these numerals is found in the South Greek counterparts. The West Greek a vocalism is the regular outcome of the syllabic nasal in PIE *-km̥ t-to- and is therefore to be regarded as an archaism. Beekes ascribes the South Greek innovation of o vocalism to the influence of the ordinal εἰκοστός, which in its turn would have taken it from τριακοστός.15 This innovation did not take place, however, in the South Greek dialect of Arcadian and in Pamphylian, in which a vocalism was retained as well.

A third phonological peculiarity of West Greek is the a vocalism in the lexeme ἱαρός ‘sacred’. This is attested in Doric, Northwest Greek, Boeotian and Pamphylian; in Elean we find, for instance, the dat. sg. ἐπιάροι ‘on the altar’ [10.9], which would be ἐφ’ ἱερῷ in South Greek. García Ramón reconstructs a Proto-Greek form *isrόs16, which was in his view modified in South Greek by replacement of the suffix with *-ero-. In West Greek, -α- was, according to him, added to the suffix under the influence of the verb ἰαίνω (< *ihan-i̯ ō < *h₁ish₂-n-i̯ oH).17

1.1.2 West Greek morphological features

Elean shares the following three morphological features with its West Greek counterparts, the first two of which indicate that Elean belongs within the West Greek group. This classification is corroborated by the third feature, which is an archaism.

12 The assibilation of -τι- to -σι- has also taken place in Lesbian, in contrast to the other Aeolic dialects. It is one of the features which Porzig (1954 : 149-51) ascribes to Ionic influence after the settlement of Aeolic migrants on Lesbos. This explanation is probable in view of the geographic proximity of Lesbos and the Aeolis to the Ionic dialect area.

13 Quotations from the inscriptions are cited according to the division used in the edition by Minon 2007, indicating the number of the inscription and the line(s) in which the particular form is found.

14 See e.g. Risch 1955 : 66.

15 Beekes 2010 : 328-9: “The original form of the second member is -κατιοι, which became -κοσιοι with regular assibilation -τι- > -σι- and analogical o after -κοντα, -κοστος.”.

16 The Pgr. form *isrόs may or may not be the result of a merger of two different PIE roots which were close in meaning, namely *h₁is-rό ‘vigorous’ and *h₁ish₂-rό ‘agile’ via Schindler’s law of laryngeal loss in the context /s_T. It is of relevance here that we are dealing with one form in Proto-Greek, from which North and South Greek both formed their respective variants.

17 García Ramón 1992 : 203-4.

1) the use of the ξ-aorist and future in dental stems 2) the local and temporal adverbs ending in -κα and -η, and 3) the nom. pl. of the article τοί, ταί.

For verbs whose present ends in -άζω or -ίζω, West Greek dialects have an aorist and future in -ξ-. This is regularly expected in velar stems such as φυλάσσω (< *pʰylak-), but in West Greek, it is also found in dental stems. Compare, for example, the Elean 3rd du. aor. ἐδικαξάταν [15.4] with the Attic form ἐδίκασα, which both derive from the Pgr. stem *dikad-. This extension of the ξ aorist and future to dental stems is also found in Arcado-Cyprian.

A second morphological criterion for separating West Greek from the other dialects are the local and temporal adverbs. West Greek uses -κα as a temporal suffix in preference of South Greek -τε, as in Elean τόκα ‘then’ [6.6], which would be τότε in classical (South) Greek. Another suffix that is typical for West Greek is the local/temporal -η, as seen in El. τɛ̃δε ‘here’ [22.20].

A third morphological feature that is typical for West Greek (and, hence, also for Elean) is the nom. pl. of the article. In Elean and in the other West Greek dialects, this is attested as τοί [10.6-7] and ταί, whereas South Greek shows οἱ and αἱ for the masculine and the feminine gender, respectively. The South Greek variant results from analogy to the singular forms ὁ, ἡ (cf. the corresponding Skt. pronouns sa (masc. sg.) and te (masc. pl.).18 Due to this analogy, the nominative forms both in the singular and the plural start in /h/. We are therefore clearly dealing with a West Greek archaism and a South Greek innovation. That this innovation is a relatively young one, is indicated by the fact that τοί and ταί, functioning as demonstratives, are still common in Homer.

1.1.3 Conclusion

As follows from the discussion in this paragraph, Elean takes part in those developments that are characteristic for the West Greek dialects. It can therefore safely be classified as a West Greek dialect. This classification is corroborated by a shared archaism, i.e. the nom. pl. of the article. In the next section, we will focus our attention on the features that Elean has in common with Northwest Greek.

1.2 Northwest Greek features

In the previous section, we have concluded that the Elean dialect belongs within West Greek, which, together with Aeolic, forms the North Greek dialect group. West Greek is, in its turn, divided into two major subgroups, dependent on how the secondary long mid vowels are treated by the individual dialects (this will be explained in paragraph 1.3). The two subgroups which together form West Greek are Northwest Greek, spoken in the area north of the Peloponnese, and Doric, spoken in the Peloponnese and on the Doric isles.19

18 Chantraine 1961 : 124-5.

19 Buck 1955 : 11-14.

In this paragraph, those features which Elean has in common with the Northwest Greek dialects will be presented. As we have seen in the introduction, there is no consensus about the question whether Elean belongs within Northwest Greek or not. Therefore, contrary to the features in the previous paragraph, one or more of these features may be of importance for our final conclusion. For this reason we will for each feature in this paragraph discuss whether it may be a potential common innovation of Elean and Northwest Greek. If so, the particular feature will be given a closer consideration in chapter 2.

1.2.1 Northwest Greek phonological features

Specifically Northwest Greek phonological features that are shared with Elean are: 1) the development of *-ӗ- to α when followed by ρ, and 2) the spelling -στ- for the outcome of *-stʰ-.

In Northwest Greek, *ӗ is lowered to α when it is followed by ρ. The same development is present in Elean. An example of this is seen in the infinitive form φάρɛ̄ ν ‘bear’ [25.5], which corresponds to classical Greek φέρειν. In Elean, however, this lowering of short ӗ is not confined to the vicinity of ρ. A series of further instances of the development *ӗ > α, in divergent phonetic contexts, is attested in the Elean inscriptions. Therefore, in paragraph 1.4, in which the specifically Elean dialect features are discussed, we will look at the instances of *ӗ > α in other phonetic contexts. On the basis of this, we will determine whether the Elean lowering of *ӗ is related to the Northwest Greek one, and whether this is a feature that deserves closer consideration.

Another peculiarity that Elean has in common with Northwest Greek is the fact that <στ> is attested as the outcome of *-stʰ-. This otherwise yields the regular outcome <σθ>, as is shown by the comparison of the 3rd sg. imp. λυσάστō ‘shall loosen’ [25.7-8] with classical Greek λυσάσθω. We are therefore dealing with an innovative change in spelling, which could unite Elean with the Northwest Greek dialect group. However, according to Méndez Dosuna, the spelling <στ> indicates a retained occlusive pronunciation of *-tʰ- after -σ-.20 In other contexts, *-tʰ- would already have developed into [θ] in Elean, as happened in many Ancient Greek dialects.21 Méndez Dosuna’s explanation is followed by Minon.22 Therefore, we are dealing with an innovation merely in spelling, representing a phonetic archaism when it comes to the pronunciation of *-tʰ- after -σ-. For this reason, it is not a valuable feature for genetic dialect classification. It will, therefore, not be further evaluated.

1.2.2 Northwest Greek morphological features

Morphological features that Elean shares with Northwest Greek are:

1) the middle participle in /-ēmenos/ 2) the consonant stem dat. pl. ending -οις

20 Méndez Dosuna 1985 : 359-60.

21 Méndez Dosuna 1985 : 360.

22 Minon 2007 : 337.

3) the consonant stem dat. pl. ending -εσσι, and 4) the acc. pl. ending -ες.

Elean (and, perhaps, Boeotian) shares with Northwest Greek that verbs whose present ends in -έω have a mid. part. in /-ēmenos/, cf. Elean dat. sg. mid. ptc. καδαλɛ̄ μενοι [10.6-7]. This is peculiar, since the ending /-ōmenos/ is expected as the result of contraction of ε and ο. A possible explanation of this long /ē/, where long /ō/ is expected, could be analogy to other forms of the verb in which ē was actually the regular product of contraction.23 This is in any case an interesting feature and deserves a closer consideration in chapter 2.

Another morphological feature that Elean has in common with the Northwest Greek group is the consonant stem dat. pl. ending. Both the endings -οις and -εσσι are attested in Elean, whereas classical Greek has merely -σι. Compare the Elean dat. pl. χρɛ̄ μάτοις [5.8] and φυγάδεσσι [30.10] with classical χρήμασι and φυγάσι, respectively. As for the origin of the endings, the first one is clearly taken from the o stems. The latter one could, according to Buck, have been extended from the s stems (which have regularly pl. -εσ- + dat. -σι) or modeled after the a/o stems. In the latter case, the dat. pl. endings -αισι, -οισι would have served as a model in order to add -σι to the consonant stem nom. pl. ending -ες.24 In the next chapter, we will try and determine whether this innovation change is an indication of dialectal relatedness between Elean and Northwest Greek.

Also in the accusative plural, there is an interesting morphologic feature both in Northwest Greek and in Elean, this being the consonant stem acc. pl. in -ες. It is for instance seen in the form χάριτερ [34.17], which corresponds to classical Greek χάριτας. The latter form displays the phonetically expected -ας from *-ns. Minon (2007 : 378) proposes as a possible source of influence the model of the nouns which have had their nom. pl. ending -εις extended to the accusative. In discussing the origin of this language tendency to level the nominative and accusative forms, she further points at the numerals, which tend to become indeclinables. This feature too will, in addition to the dat. pl. endings and the middle participle in/-ēmenos/, be evaluated in the next chapter.

1.2.3 Conclusion

In this section, it has become clear that Elean and Northwest Greek share certain features, i.e. all four morphological ones discussed above, which are potential common innovations. For this reason, these features will undergo a closer consideration in chapter 2 in order to determine whether Elean belongs within the Northwest Greek dialect group. The spelling <στ> for the outcome of *-stʰ- does not, as we concluded, necessarily represent a sound change, and will therefore not be given a closer look. The lowering of *-ӗ- to α will first be discussed in its more specific operation contexts in Elean. After that, we will decide whether this feature too will be evaluated in chapter 2.

23 Buck 1955 : 124.

24 Buck 1955 : 89.

1.3 Features in common with various other dialects

There is a wide range of features that Elean does not specifically share with West Greek or Northwest Greek, but with various dialects inside and outside these two dialect groups. To these features, this paragraph is dedicated. They will be separated into phonological and morphological features and be discussed in their respective subparagraphs.

1.3.1 Phonological features in common with various other dialects

The phonological features which Elean shares with various other dialects are the following:

1) the merger of *ō and ō₂ into one single phoneme /ō/ 2) the loss of initial aspiration (psilosis) 3) the development *-zd-, -gi̯ -, -di̯ - > -δ(δ)- 4) the development *-rs- > -ρρ- 5) word-final rhotacism 6) the late development *-s- > -h-, -Ø- intervocalically 7) the retention of initial digamma 8) the loss of intervocalic i̯ , and 9) the development *-VnsC > -αις, -οις.

The development of the long mid vowels ē and ō is an isogloss that cuts right through the West Greek dialect group. In some West Greek dialects, originally long *ē and *ō do not merge with the secondary long ē and ō which resulted from compensatory lengthening. The same is seen in classical Attic and therefore, these dialects are referred to as doris mitior. In the Doric proper dialects however, originally long *ē and *ō merge with the secondary ones. These dialects are therefore called doris severior.25 Elean shows a middle position between this subdivision within West Greek: it merges *ō and ō₂, but it keeps *ē and ē₂ separate.26 This is for instance shown by the Elean nom. sg. βōλά ‘council’[4.2] (= classical Greek βουλή) and ὀπτṓ ‘eight’ [25.4] (= classical Greek ὀκτώ), showing one phoneme /ō/ whereas Attic has <ου> for secondary long ō and <ω> for originally long *ō. In the next chapter, we will investigate what we can deduct from this unique long vowel system with respect to the position of the Elean dialect.

Another phonetic peculiarity is the absence of initial aspiration, also known as psilosis, in Elean. This is most clearly visible in cases where a voiceless stop is followed by a word which would in non- psilotic dialects start with an aspirate. The fact that in Elean, the sequence of sounds is written as a simple voiceless stop π, τ, κ, rather than an aspirated stop φ, θ, χ, indicates that the initial h- has dropped. For example in the phrase ἐπιάροι ‘on the altar’ [10.9], the π indicates that ιάροι is to be read as ἰάροι, rather than ἱάροι, as in the classical Greek counterpart of this phrase ἐφ’ ἱερῷ. Psilosis is shared with East Ionic, Lesbian, Cyprian and Cretan. These are all dialects from different dialect groups. Within these groups, the other dialects do not show psilosis. Therefore, it is probably a young feature that may have independently operated in different dialect groups, and is for that reason not relevant for our investigation.

25 Méndez Dosuna 2007 : 451.

26 Méndez Dosuna 2007 : 452.

A feature no less notable are the spellings <δ>, <δδ>, <ττ> for the outcome of Proto-Greek *-zd-, -gi̯ -, -di̯ -, all of which yielded <ζ> in classical Attic.27 Take for example the 3rd pres. opt. δικάδοι [4.2], which is in classical Greek δικάζοι. The different spellings make it hard to determine the exact underlying sound value. It is, furthermore, not certain whether these spellings actually represent a sound change, for it is known that in Elean, <ζ> was used in order to spell the outcome of *d (as will be discussed in paragraph 1.4.1). Therefore, it is not improbable that an alternative spelling was sought for the outcome of *-zd-, -gi̯ -, -di̯ -.28 In this case, we would be dealing with a merely graphic feature, which does not represent an actual sound change. Due to this uncertainty, this feature will not be evaluated in the next chapter.

The present infinitive form θαρρɛν̃ ‘dare’ [20.1] indicates that in Elean, the sequence *-rs- developed into -ρρ-. Classical Attic has the same development, as seen in the cognate form θαρρɛῖν, but Ionic θαρσɛῖν displays the original sequence of sounds. Arcadian and Theran show the same development as Attic and Elean. It is, for that reason, just like psilosis, probably a late feature which may well have occurred in different dialects independently. Therefore, it will not be evaluated in chapter 2.

A similar case is rhotacism, as is seen in the nom. sg. personal pronoun ὄρτιρ [6.6] (= classical Greek ὅστις). That rhotacism is a relatively late development in Elean appears from the fact that only in the younger inscriptions it is written consistently. The older inscriptions show a hesitation between -ς and -ρ word-finally. Since rhotacism is obviously a recent development, it will not be taken into further consideration.

Τhe secondary loss of the retained intervocalic -σ- in aorist forms is also young. In Attic, intervocalic -σ- was restored (or simply kept) in the aorist forms. In Elean, however, a second lenition of intervocalic -σ- took place, yielding forms such as ἀδεαλτώhαιε [30.12]. The aspiration that resulted was probably sufficiently distinctive for the form to be recognised as an aorist. Since older inscriptions still show the -σ- in the aorist, as appears from for instance the form καθύσας [3.2], this is clearly a late development and, therefore, not relevant for our question.

In Elean, similarly to -σ-, intervocalic -ι- dropped in the sequences -αι-, -ει-, -οι-, -υι- when followed by a vowel. This is seen in the 3rd pres. opt. ἔα ‘be’ [1.4] (= classical Greek εἴη). This development is also found in later Attic, in which case it only took place before a front vowel. This is shown by the 3rd sg. fut. form ποήσει ‘will make’ from ποιήσει. In Elean, however, the spellings are not consistent, for, as opposed to ἔα [1.4], we also come across ἔιɛ̄ [4.1]. Moreover, this development is also found in a range of dialects from the Arcado-Cyprian, Doric, and Aeolic group. This makes it probable that this is a late, independent development. Neither the loss of -σ- nor that of -ι- will therefore be evaluated in chapter 2.

The retention of initial ϝ- is a clear archaism in Elean, as it is in Laconian, Cretan, and Argolic. It is found in for instance ϝέκαστον [9.3] ‘each’ (= classical Greek ἕκαστον). Due to its clearly archaic character, this feature too will play no key role in our investigation.

27 Buck 1955 : 71.

28 Minon 1998 : 181-2.

Finally, the outcome of the acc. pl. ending *-ns needs to be discussed. In Elean, as in Lesbian, this turns up spelled as -αις, -οις, cf. μναῖς [13.4], which corresponds to classical Greek μνᾶς. Whether this spelling actually represents a particular sound change is actually uncertain. The -ι- may as well be an indication of vowel length, cf. the Attic spelling <ει> for /ē/. This would well be in accordance with the fact that acc. pl. in /-ās/ and /-ōs/ are also attested for Elean, cf. δαρχνάς [5.7] ‘drachmas’ and ἰα[ρ]ομάορ [14.6]. It appears we are dealing with an ending not fundamentally different from that of the other Greek dialects. Therefore, this feature too will not be taken into further evaluation.

1.3.2 Morphological features in common with various other dialects

The following range of morphological features are shared with various individual dialects both inside and outside the West Greek group:

1) the retention of the nom. sg. ending -α 2) the dat. sg. ending -οι 3) the uncontracted nom. pl. form βασιλάες 4) the superlative ἄσιστα 5) the attachment of the deictic suffix -ί to the article 6) the deictic gen. pl. neut. ταύτōν analogous to ταῦτα 7) the preposition ὑπά analogous to κατά 8) perfect reduplication versus plain vowel 9) the imperative in (-ντōν), -ōστōν 10) the 3rd sg. subj. in -ᾱ < *-ē 11) the aor. subj. in -σᾱ- 12) the 3rd sg. σ-aor. opt. in -σειε, -hαιε and, 13) the μι-forms in the opt. of contracted verbs.

None of these will be of relevance for our question. In some cases, the particular development is a late one, such as the gen. pl. neut. deictic pronoun ταύτōν [22.5] (= classical Greek τούτων). Here the original form *τούτōν has been replaced by ταύτōν by means of analogy to the nom. and acc. forms. Another example of a recent analogical development is the preposition ὐπά for ὐπό. Apart from Elean, this variant is apart from Elean found in Lesbian, Boeotian, and Locrian. An Elean example is seen in the dat. pl. ὐπαδυγίοις ‘mules’ [9.9]. In this case, the prepositions κατά and ἀνά are the most probable source of analogical influence. Another instance of analogy is seen in the superlative ἄσιστα [30.9] ‘nearest’. This form is found in Elean and Laconian; elsewhere the regular variant ἄγχιστα is attested. Most probably, the comp. *ἆσσον, which is the regular outcome of Pgr. *ankʰ-i̯ on-, has here served as a model. The single sigma in the superlative ἄσιστα remains of uncertain origin.

Some other morphological peculiarities of Elean are no instances of some analogical development, but are nonetheless young developments. Therefore, these too will be not be selected for evaluation in the next chapter. This is for instance the case in the dat. sg. art. τοῖ [28.1], which has a short vowel diphthong, whereas its classical Greek counterpart τῷ has a long one. The form could originally be the locative form, in which case the difference between Elean and Attic is the result of selection. More probably, however, the short ŏ has resulted from the shortening of final diphthongs, which is not an uncommon process in the Indo-European languages. In neither case, however, the feature would deserve further attention in the next chapter.

The deictic suffix -ι is another case where we are dealing with a selection out of a number of options from the Proto-Greek stock. This suffix is attached to the article in order to create a deictic pronoun: Elean has τοΐ [10.3], whereas classical Attic prefers the ‘proper’ deictic pronoun τοῦτο. A comparable form οὑτοσί is, however, also attested in Attic. A further example of selection is seen in the perfect participle ɛ̄ γραμένοι [10.10], in which Elean and Cretan show a plain vowel as a marker of the perfect tense. The classical Greek cognate form γεγραμένοι in its turn shows reduplication. Similarly, the ‘Aeolic inflexion’ in the optative of contracted verbs συλαίɛ̄ [12.6] is shown by Risch to be a case of selection.29

An archaism, which is too no strong indication of dialectal relatedness, is for instance the retention of -α as a nom. sg. masc. ending. Most dialects have in these cases added -ς to the word, in order to typify its masculine gender (cf. Attic ποιητής). Elean and Boeotian, however, have retained -α. An example of this is seen in for instance τελεστά [10.8-9]. The uncontracted nom. pl. form βασιλάες [20.3] (= classical Greek βασιλεῖς) is anοther example of an archaism, as is the 3rd sg. subj. ἐκπέμπᾱ [30.11]. Here, Elean, Laconian, and Aeolic lack the addition of -ι to the verb form, which happened in the other dialects under the influence of the indicative, cf. Attic ἐκπέμπῃ. The fact that Old Ionic still has a form πίɛ̄ σι30 ‘drinks’ (3rd sg. aor. subj. of πίνω) indicates that the addition of ι is a late development.

1.3.3 Conclusion

As we have seen, most of the features we discussed in this section are not relevant for our question. The majority of them are possibly or probably recent developments, some are archaisms, others are selections. Neither of these are strong indications of dialectal relatedness. In some cases, it is not certain whether a certain spelling represents an actual sound change, or whether we are simply dealing with a spelling variant.

The only feature discussed in this chapter which may be valuable for our investigation is the merger of *ō and ō₂. This feature would in itself point to a classification of Elean under the Doric dialect group. However, in paragraph 1.4, we will learn that Elean does, on the other hand, not merge *ē and ē₂, which in its turn suggests a classification under the Northwest Greek group. The development of the Elean long vowel system will therefore be given a loser consideration in the next chapter.

1.4 Specifically Elean features

The last paragraph of this chapter is reserved for the features that are typical for Elean. These are the features that give Elean its individual character respective to the other Northwest Greek and West Greek dialects. Since these features are not shared with any other dialect, they cannot be indications of dialectal relatedness between Elean and some other dialect. They may, however, give us an answer to the question as to when Elean separated itself from its West Greek ancestral branch.

29 Risch 1955 : 71-2.

30 Miller 2013 : 98-9.

1.4.1 Specifically Elean phonological features

Again, the phonological and morphological features are presented in separate subparagraphs. The typically Elean phonological features are:

1) the development of *ē to /ǟ/ 2) the lowering of *ӗ to α 3) the lowering of *i to ε 4) the spelling of <ζ> for *d 5) the development *-skʰ- > -σκ-, and 6) the development *-stʰ - > -σσ-.

As mentioned in paragraph 1.3, Elean takes a middle position between Doris mitior and Doris severior, for it merges primary and secondary long ō, but keeps primary and secondary long ē apart. Primary long *ē is spelled sometimes <ε>, sometimes <α>, whereas ē₂ is consistently spelled <ε>.31 Hence, we find both ϝράτρα ‘decree’ [4.2] and ϝρɛ̄ ̄́τ[ρα] [23.3]. The innovative character of this sound change appears from its absence in other dialects. Moreover, it could very well be an old one, as has been argued by Bartonĕk.32 In search of an old innovative change of Elean, this is a feature that deserves further consideration in the next chapter.

A similar process has taken place in the short vowel system. Here, we find lowering of *ĕ to α when an ρ follows. As seen in paragraph 1.2, this is a general Northwest Greek feature, but in Elean, this lowering operates in other environments as well. In the infinitive γνōμαν [14.6] (= classical Greek γνῶναι, but with the West Greek inf. ending -μεν) for instance, we see the same development before ν, but there are also examples from again different phonetic contexts. This raises the question whether in Elean, the sound change is actually context-dependent, and whether it is related to the lowering of *ĕ before ρ in Northwest Greek. Its similarity with the development of *ē to /ǟ/ raises the possibility of interrelatedness between these two sound changes. Therefore, this development could as well be an old innovation. For this reason, the lowering of *ĕ to α will undergo further consideration in the next chapter as well.

In Elean, lowering is not only found for *e, but also for *i, as is shown by the nom. sg. form πόλερ [34.16] (= classical Greek πόλις). In this case the lowering must be a younger development, whose terminus post quem is the rhotacism of final -ς. In case of relatedness between this change and the lowering of *e, the latter one must also be a young development. This question is, however, a matter to be discussed in chapter 2.

Another uniquely Elean feature is the so-called zetacism, i.e. the spelling <ζ> for the outcome of *d. An example of this is the nom. sg. ζίκα [1.1] (= classical Greek δίκη). It has been suggested that this spelling represents a spirantisation or palatalisation of *d.33 However, Minon gives preference to the explanation that <ζ> started to become used as an alternative spelling for the outcome of *d after

31 Minon 2007 : 287.

32 Bartonĕk 1972 : 209-10.

33 Minon 1998 : 194 - 203.

*-zd-, -gi̯ -, -di̯ - developed into /d/ in Elean (cf. paragraph 1.3.1).34 If we follow her explanation, the <ζ> spelling is a merely graphic feature, representing simply plain /d/.

In Elean, the spelling of a voiced stop for a Proto-Greek aspirate after σ is not confined to the dentals, as it is in Northwest Greek. In Elean, a similar process turns up in the velar series, as shown by for instance the 3rd sg. opt. πάσκοι [20.8] (= classical Greek πάσχοι). As is the case with the spelling <στ> for the outcome of *-stʰ-, it is not certain whether a particular sound change is indicated by this spelling; according to Thevenot-Warelle, it represents the retention of [kʰ] after -s-, whereas otherwise, *kʰ underwent an early process of spirantisation.35 It will, therefore, be given no further evaluation in the next chapter.

The final phonological peculiarity of Elean that will be discussed is the development of *-stʰ- to -σσ-, as seen in the middle infinitive ἀποδόσσαι [30.9] (= classical Greek ἀποδόσθαι). In accordance with the outcome of *-tʰ- and *-kʰ-, a spelling <στ> would here be expected. A possible scenario is that the aspirate character of /th/ was retained after the phonetically proximate -s-, after which it spirantised and subsequently assimilated to the -s-.36 The development is in any case limited to the younger inscriptions of Elean, and is therefore not one that requires a more thorough discussion in the next chapter.

1.4.2 Specifically Elean morphological features

The following are the morphological features that are typical for Elean, most of which result from recent analogical processes:

1) the gen./dat. dual ending -οιοις 2) the use of μεύς ‘month’ under the influence of Ζεύς 3) the use of ὀπτώ analogous to ἑπτά 4) the present in -είω, and 5) the 3rd sg. imp. ἤστω.

Three of these, the forms μεύς for ‘month’, ὀπτώ for ‘eight’, and the 3rd sg. imp. ἤστω, can easily be ascribed to relatively recent analogical developments, and are therefore not relevant for our research. In the first case, *μɛ̄ ς is expected as the regular nom. sg. form. The form μεύς [22.15], with its genitive μɛ̄ νός [7.2] probably came into being after the analogy of Ζεύς, whose plural form Ζᾶνες was used to denote Zeus statues.37 Ὀπτώ ‘eight’ [25.4] most probably owes its π to its neighbour ἑπτά ‘seven’. The long vowel in the 3rd sg. imp. ἤστω may have extended from the infinitive, which competed with the imperative in order to express a command.38

34 Minon 2007 : 332-3.

35 Thevenot-Warelle 1988 : 103.

36 Minon 2007 : 339.

37 Minon 2007 : 290.

38 Minon 2007 : 390.

One clear archaism from this list of specifically Elean morphological features is the present in -είω, which is for instance seen in the dat. pl. prs. mid. ptc. πολιτειομένοιρ [28.5-6] (= classical Greek πολιτευομένοις). As appears from the Attic form, classical Greek here has the suffix -εύω, as do the other dialects. This variant, however, probably results from analogical influence from other forms of the verb where -ευ- was regular, such as the aorist in -ευσα, the future in -ευσω and the perfect in -ευκα.39 As is the case with the other archaisms discussed so far, this one will not be evaluated in the next chapter.

Finally, the remarkable Elean gen./dat. dual ending remains to be discussed. It is found as -οιοις and -οιοιρ in o stem nouns, as for instance in καὐτοίοιρ [19.3-4] (= καὶ αὐτοῖν). Most other dialects have -οιιν (or -οιν by contraction), apart from Arcadian, which has -οιυν. According to Minon, this form originates from the regular gen./dat. dual ending -οι- (< *-οι-ι) with the dat. pl. ending -οις added to it.40 We are here therefore dealing with a selection out of the endings which were available in Proto- Greek. This feature too will therefore not be discussed more thoroughly in chapter 2.

1.4.3 Conclusion

As we learned in this paragraph, none of the morphological features which are specifically Elean are of relevance for our question. On the phonological side, however, we have come across three developments in the vowel system which are potentially old. These three features, i.e. the development of *ē to /ǟ/, the lowering of *ӗ to α, and the lowering of *i to ε, will, therefore, be taken to the next chapter in order to undergo further evaluation.

1.5 General conclusion

In this paragraph, I will sum up all features which we discussed in this chapter and which were decided to be taken to chapter 2 for closer consideration. These features are the following:

1) the middle participle in /-ēmenos/ 2) the consonant stem dat. pl. endings -οις and -εσσι 3) the acc. pl. ending -ες 4) the development of *ē to /ǟ/ 5) the lowering of *ӗ to α, and 6) the lowering of *i to ε.

For all of these features, we will try and determine whether they are old innovative changes, and, if so, what they tell us about how Elean fits within the West Greek dialect group.

39 Minon 2007 : 397.

40 Minon 2007 : 372-3.

2 Evaluation

In this chapter, I will select and discuss thoroughly some remarkable Elean features which may give us an indication on the position of Elean relative to the other West Greek dialects. The focus will - whenever possible - be on internal reconstruction and genetic comparison, rather than secondary influence from neighbouring dialects and/or substratum. Three morphological features will be discussed, as well as three phonological ones. The morphological ones are the accusative plural in -ες, the dative plural in -οις and -εσσι and the middle participle in /-ēmenos/. The phonological ones are the development *ē > ǟ, the development *ӗ > α, and the development *i > ε.

2.1 The consonant stem accusative plural in -ες As mentioned in paragraph 1.2.2, Elean and Locrian inscriptions show a consonant stem accusative plural ending in -ες rather than the expected -ας, as seen in the following forms: ὀμόσαντες [22.11], πλείονερ [34.8], χάριτερ [34.17]. The development is attested in inscriptions from the fifth century B.C. onwards. Apparently, since -ες is the regular nominative plural ending for the consonant stem nouns, these forms result from levelling between the nominative and accusative endings.

This is a feature which has parallels in other Ancient Greek dialects. In Attic, for example, the distinction between the nom. and acc. pl. forms is known to have been leveled out in certain paradigms.41 Examples are the nouns of the type ἀληθής (nom./acc. pl. ἀληθεῖς) and βασιλεύς (nom./acc. pl. βασιλεῖς), the comparative adjective (nom./acc. pl. κρείττους), and the nouns πόλις and πῆχυς (nom./acc. pl. πόλεις, πήχεις).

In the case of πόλις and πῆχυς, however, the analogy between the nominative and accusative forms need not be the result of paradigmatic leveling, as Chantraine (1961 : 88-9) points out. According to him, a secondary stem πολε-, πᾱχε-, with generalised e vocalism, could have been formed in these paradigms. To this stem, the regular endings *-es and *-ns were added in the nominative and accusative respectively. After the completion of the compensatory lengthening and contraction of like vowels, the forms *πολενς, *πᾱχενς created in this way would, in Attic, end in -ɛις, i.e. the attested πόλεις, πήχεις.42 The thereby created forms *πολενς, *πᾱχενς would, after the completion of the compensatory lengthening and contraction of like vowels, in Attic, end in -ɛις, i.e. the attested πόλεις, πήχεις.43 This ‘merger’ of the nom. and acc. pl. forms could then have served as a model for other third declension nouns.

Let us at this point investigate whether such a model would work for Elean as it would for Attic. In other words: whether the Elean reflexes of *e-ns are the same as those of *e-es, as they are in Attic. As we have seen in par. 1.3.1, in Elean, both the compensatory lengthening of *e as well as the contraction of e + e result in the new long /ē/ phoneme, which is consistently spelled <ε>. Therefore, the model as described by Chantraine would work for Elean as well.

It has been suggested by Méndez Dosuna that the use of the nom. ending in order to denote the acc. case has spread from the numerals, where ‘four’ may have been influenced by the higher ordinals ‘five’ and ‘six’. These were indeclinable and did, therefore, not distinguish between their nominative and accusative forms.44 According to Chantraine, influence from the preceding number ‘three’, whose

41 Chantraine 1961 : 60.

42 Chantraine 1961 : 88-9.

43 Chantraine 1961 : 88-9.

44 Méndez Dosuna 1985 : 465.

accusative was in certain dialects equal to the nominative, is also to be reckoned with.45 Following these examples, the originally nominative form τέτορες could have been used as an accusative as well.

However, it is very unlikely that influence from the numerals is strong enough to affect the declension of entire noun classes, in this case the third declension. Therefore, I prefer to assume that the ‘regular’ merger, as described by Chaintraine as a possible source for the Attic homophony between nom. and acc. pl., also took place in Elean and Locrian. Here, the addition of the endings *-es and *-ns to the secondary stem πολε-, πᾱχε- would, after the completion of the compensatory lengthening and contraction of like vowels, lead to nom. and acc. forms both ending in -ɛ̄ ς, i.e. πόλɛ̄ ς, πᾱχɛ̄ ς. These leveled nominative and accusative plural forms could then have influenced those of the other consonant stem nouns. The indistinct nom. and acc. forms of the numerals ‘five’ and ‘six’ would, in this event, merely have served as an extra backing for this tendency, rather than being the cause of it.46

So far, we have discussed the possible origin of the consonant stem accusative plural ending in -ες. It becomes apparent from our discussion that the extension of -ες to the accusative is a development which could take place in different dialects independently. The fact, however, that in Elean and Locrian, this feature is already present in the older inscriptions, whereas in other dialects, it appears only as early as the third century B.C.47, makes it a very interesting common innovation of these two dialects. It appears that Elean and Locrian have invented this feature long before the other dialects started to show the same tendency, so that in the case of these two dialects, it cannot be ascribed, as Méndez Dosuna does48, to drift. Rather, it is an old shared innovation of Elean and Locrian, dating from the early 5th century B.C. at last, but probably from much earlier. It is, therefore, a strong indication of relatedness between these dialects.

2.2 The consonant stem dative plural endings -οις and -εσσι We have learnt from our discussion in par. 1.2.2 that dative plural forms in -οις and -εσσι are attested in various consonant stems in numerous West Greek and Aeolic dialects. It is a difficult task to recognise a specific distribution between the different endings, since the material is scarce and in some dialects, different endings occur within the same declension group or even within the same lexeme. For example, we come across [γ]ραμματεσσι and χρειματεσσι, but also χρɛ̄ μασιν in Thessalian.49 In Elean, the ‘Aeolic’ dative plural is attested in the forms Θεσπιέσσι [15.5] and φυγάδεσσι [30.10]50. The -οις ending is found in the forms χρɛ̄ μάτοις [5.8], ἀγώνοις [1.26].

45 Chantraine 1961 : 60.

46 In some dialects, this leveling would also have taken place in 'three' and 'four' after the model of ‘five’ and ‘six’. In Elean, this is perhaps visible in [τετο]ρες μνᾶς [21.3-4], if -ρες is indeed to be read as ‘τετορες’.

47 Méndez Dosuna 1985 : 466.

48 Méndez Dosuna 1985 : 466.

49 García Ramón 1990 : 136-7.

50 Perhaps also in ΜαντινΕΣι, if the single Σ represents a geminate and the form is to be read as /Mantinessi/ (García Ramón 1990: 151). The spelling would then be due to the fact that in this inscription, the notation of geminates is very inconsistent.

Since the -εσσι ending is present in Boeotian, Thessalian and Lesbian throughout, it is often referred to as an Aeolism and has been reason for some scholars to assume language contact with Aeolic by speakers of those West Greek dialects in which -εσσι datives are also found. García Ramón (1990), however, indicates how this formation could have arisen within different individual dialects on the basis of proportional analogy. If we follow his argument, the dative in -εσσι does not necessarily provide an indication of dialectal relatedness or language contact.51

As for the motivation for the replacement of the original consonant stem dat. pl. ending -σι, García Ramón points to the nouns whose stems end in -n-, in -nt-, or in a dental. In those cases, the regular dat. pl. form ending in -σι would, in combination with these suffixes, have undergone sound changes, as happened in Ionic-Attic. For example, the respective dat. pl. forms of ἀγων- ‘match’ and φυγαδ- ‘exile’ are ἀγῶσι and φυγάσι in Attic. In these forms, the noun stem has been mutilated, whereas in all other forms it remained intact (apart from the nom. sg. φυγάς < *φυγάδ-ς). This irregularity was mended by rebuilding these dative plurals: the endings -οις or -εσσι were used for the dative of nouns whose stems ended in -n-, -nt-, or a dental. It is indeed in words belonging to these noun classes that the -οις and -εσσι endings are attested.52

The following model, which García Ramón quotes from Wackernagel, could be the basis from which the -οις and -εσσι endings have spread:

λόγων : λόγοις – ἀγώνων : ἀγώνοις λύκοι : λύκοισι – θῆρες : θήρεσσι53

That is to say: ἀγώνοις has been built taking the -οις ending from the o stem nouns and attaching it to its stem ἀγων-. This stem is thereby kept intact, providing a more transparent paradigm for the n stems. The consonant stems thus show a tendency to merge their plural endings with those of the vowel stems in the weak cases, i.e. the genitive and the dative.

On the basis of the model λύκοι : λύκοισι, in which the dat. ending -σι is attached directly to the nominative plural form λύκοι, the same formation was extended to ἀγών, i.e. ἀγῶνες + -σι > ἀγώνεσσι. Since the word stem is still ἀγων-, -εσσι could here be reinterpreted as a dat. pl. ending. Thence, it could have become used in other cases where the original ending -σι would lead to stem modifications, but an ending starting with a vowel would not.

Of course, the ἀγώνεσσι model would originally only work with masculine and feminine consonant stem nouns, whose nom. pl. ending is -ες, which is a prerequisite for this analogical model to work. Only secondarily it would have spread to the neuter nouns, whose nom. pl. ending is -α.54

51 Cf. García Ramón 1990 : 144-5: “… wenn es tatsächlich einen und nicht verschiedene [Ausgangspunkte] gibt (…). Es ist hingegen m.E. sicher, dass sich der Dat. Pl. auf -εσσι (…) in jedem beliebigen aiolischen oder (nord)westgriechischen Dialektgebiet (…) entfalten und zusammen mit anderen Allomorphen bestehen konnte.”.

52 García Ramón 1990 : 135.

53 García Ramón 1990 : 142.

54 García Ramón 1990 : 146.

To the first model, García Ramón adds the article and the adjective as possible sources of extra influence:

τὰ δῶρα : τὰ (ἴδια) χρήματα τῶν δώρων : τῶν (ἰδίων) χρημάτων τοῖς δώροις : τοῖς (ἰδίοις) x; x  χρημάτοις55

It is in these consonant stem nouns belonging to the -n-, -nt-, and dental stems where a new dat. pl. ending most probably has arisen. From these, it has spread to the rest of the consonant stem nouns.

In my view, García Ramón makes sufficiently clear how the dative plurals in -οις and -εσσι could have arisen in different dialects independently.56 It is, therefore, not necessarily an indication of genetic relatedness between the respective dialects in which it is attested.

2.3 The middle participle in /-ēmenos/ As seen in par. 1.2.2, middle participles ending in /-ēmenos/ are attested in inscriptions from Phocis, Locris, Elis, and Boeotia. Examples from Elean are καδαλɛ̄ μενοι [10.6-7] and καζαλɛ̄ μενον [22.19]. In West Locrian, we find, for instance, ενκαλειμενōι. Delphian has, among other instances, καλείμενος. One form, πολημενω, is attested in Epizephyrian Locrian, and in Boeotian, we come across a form such as αφικνεμενων.57 As far as the Boeotian attestations are concerned, however, Hock (1971) has questioned whether they are genuine instances of the participle in /-ēmenos/ in this dialect.58 We will discuss this matter at the end of this section.

The participle forms in /-ēmenos/ are peculiar, since, as we have seen above, /-ōmenos/ would be the expected regular outcome. Méndez Dosuna (1985 : 499-506) discusses a range of possible explanations. A connection with the athematic conjugation of contracted verbs, which is found in Aeolic, has been suggested. In this case, however, **-ημενος would be expected in all Northwest Greek dialects rather than the actually attested -ειμενος.59 60 The origin of this ending must, therefore, be sought elsewhere.

55 García Ramón 1990 : 146.

56 García Ramón 1990 : 155.

57 Méndez Dosuna 1985 : 491-2.

58 Méndez Dosuna 1985 : 492-4.

59 Méndez Dosuna 1985 : 499.

60 The -η- in Ep. Locr. πολημενω is due to the fact that the more recent Locrian inscriptions, in which this form is attested, indiscriminately use <η>, <ω> for both ancient and secondary long /ē/,/ō/ respectively. (Méndez Dosuna 1985 : 76).

A purely phonetic explanation has been suggested by Hock (1971). According to him, the sequence -eo(C)e- developed into -ee(C)e- by means of vowel assimilation. Thus, the 1st pl. forms ending in -εομες, -εομεθα would have had their sequence -εομε- developed into -εεμε-, which would be supported by the 2nd. pl. ending -εεσθε. The model would then have extended itself to the 3rd person, yielding -εενται, -εεντο.61

However, Méndez Dosuna indicates that such a development is improbable, since in the cases above, the vowels that undergo assimilation are accented. This is an unfavourable condition for a vowel to undergo change. Secondly, one would not expect an -o- to lose its labial character and develop into ε when followed by a labial μ in the 1st plural. Thirdly, in cases of assimilation in which -e- and -o- play a role, the o timbre is usually the dominant factor, making it unlikely for the -o- to develop into ε.62

Therefore, Méndez Dosuna rather assumes a paradigmatic leveling within the conjugation of the verbs in -εω. This would serve to eliminate the difference in the number of syllables between the forms where vowels of the same quality are contracted (e.g. ποιεισθε < *ποιεεσθε) and those forms where hiatus between ε and ο is kept (e.g. ποιεομεθα, ποιεονται63). This interpretation is accepted by Minon.64 One may, however, wonder why a different number of syllables in different forms from the same paradigm might be inconvenient for a speaker in any way. Therefore, I doubt whether this is the correct explanation of this phenomenon, but so far, it is the most plausible one that has been suggested. In any case, the difficulty to explain this innovation makes it improbable that it has operated in multiple dialects independently.

Moreover, it is doubtful whether the Boeotian forms are genuine instances of this phenomenon. In Boeotia, three forms which contain an alleged participle in /-ēmenos/ are attested. Apart from the aforementioned αφικνεμενων, we find δειμενοις and ἀδικείμενος.

First of all, it is worth mentioning that the form αφικνεμενων is not attested in Boeotia itself, but in an inscription from Oropus. This participle form is, according to Buck, to be ascribed to Boeotian influence.65 Méndez Dosuna, however, argues that the Boeotian linguistic influence on Oropus starts only at the end of the fourth century B.C., whereas the inscription in which it is attested dates from 386 B.C. This is indicated by the fact that the inscriptions from Oropus that date from this age still display local linguistic peculiarities, which would, under Boeotian influence, have been replaced.66

61 Méndez Dosuna 1985 : 501.

62 Méndez Dosuna 1985 : 502.

63 Méndez Dosuna 1985 : 502-3.

64 Cf. Minon 2007 : 395-6: “Pour éviter des formes comportant deux syllabes de plus que celles dans lesquelles la mise en contact de /e/ + /e/ entraînait une contraction, dont le e paradigmatique, atone, risquait de ne plus être audible, on aurait substitué à -εόμενοι la forme en -ɛ̄ μενοι, à la voyelle longe analogique des formes contractées.”.

65 Buck 1955 : 124.

66 Méndez Dosuna 1985 : 494-5.

Secondly, the form δειμενοις is suspect. Since, in Boeotian, the verb δέω follows the inflection of the -i̯ e-/-i̯ o- verbs rather than that of the contracted verbs, *δειομενοις is expected as a middle participle.67 Indeed, according to Méndez Dosuna, δειμενοις is to be read as δει[o]μενοις. The -ει- spelling he ascribes to a more open pronunciation of short ε before another vowel.68

Thirdly, the form ἀδικείμενος remains to be explained. It is part of a Theban’s speech uttered in the play Acarnienses by Aristophanes. It is known, however, that Aristophanes sometimes falsely ascribes certain dialectal forms to people from certain regions.69 Therefore, this form too is not a reliable instance of the /-ēmenos/ participle in Boeotian.

Summarised, it appears that only in Phocian, Locrian, and Elean, we find genuine instances of this phenomenon. Let us now try and get a more precise view on when its origin is to be dated.

The terminus ante quem for this innovative change is, according to Méndez Dosuna, about 675 B.C., i.e. the foundation of Epizephyrian Locris. For, as we have seen above, an instance of this feature is also attested in this Locrian colony, and it is improbable for a change to arise both in a colony and in its motherland independently.70

Méndez Dosuna proposes the Mycenaean era as a terminus post quem, for in the Linear B tablets, jod is still spelled in verbs belonging to the -εi̯ ω conjugation, e.g. to-ro-qe-jo-me-no /tropei̯ omenos/. This indicates that in the verbs of this type, vowel contraction could not have taken place yet.71 It cannot even be excluded that in West Greek, the loss of intervocalid yod and the subsequent vowel contraction dates from an earlier period than in Mycenaean. This would put this change even further back in time.

We conclude that we are here dealing with a relatively early (at least prior to 675 B.C.) innovation, which Elis has in common with Phocis and Locris. Since, as we have seen above, it is one which has most probably not operated independently in different dialects, it is apparently a common innovation of these three dialects and, for that reason, a strong indication of genetic relatedness between them.

2.4 The development *ē > ǟ As learnt from par. 1.1.3α above, Elean merges the inherited long *ō with the ō₂ which resulted from compensatory lengthening and contraction of like vowels. Inherited long *ē and the younger ē₂ are, however, kept distinct. While *ē is spelled sometimes <α>, sometimes <ε> in Elean inscriptions, probably representing a phoneme /ǟ/, ē₂ is spelled <ε> consistently.72

67 Hock 1971 : 188.

68 Méndez Dosuna 1985 : 493-4.

69 Méndez Dosuna 1985 : 494.

70 Méndez Dosuna 1985 : 505-6.

71 Méndez Dosuna 1985 : 505-6.

72 Méndez Dosuna 1980 : 183.

In paragraph 1.3.1, we have seen that the dialects referred to as doris mitior are those that keep the long mid vowels which result from compensatory lengthening separate from the mid vowels which were originally long. This leads to a system of seven long vowels, similar to that of Attic. Doris severior are those dialects that merge the products of compensatory lengthening with the originally long mid vowels, which creates a system of five long vowels. Since, as mentioned above, Elean merges long ō, but distinguishes inherited long *ē and secondary long ē₂73, it ends up with a system of six long vowels. It thereby takes a middle position between doris mitior and doris severior.

The uniqueness of this vowel system has been reason for Bartonӗk (1972) to classify Elean as a separate dialect group within West Greek.74 This idea is, however, disputed by Méndez Dosuna. As he remarks, for a dialect to be considered separate from its mother dialect group, it must introduce a change which is not shared by any other dialect within the group. Furthermore, the dialect may not already have been separated from its mother dialect group by another, preceding change.75

According to Bartonӗk, the sound change *ē > ǟ separates Elean from the rest of the West Greek dialects. His argument is based on relative chronology. As we have seen above, the treatment of the secondary long vowels ē₂, ō₂ causes the West Greek dialects to split up into doris mitior and doris severior. Therefore, in order for Elean to be a separate descendant from West Greek, the change *ē > ǟ must have preceded the change that caused West Greek to split up. This leads Bartonӗk to propose the following chronological order of developments:

1. Elean *ē > ǟ; 2. Compensatory lengthening of short ĕ, ŏ; 3. Merger of the newly acquired ē₂, ō₂ with original *ē, *ō in doris severior.

In this order of events, the development *ē > ǟ separates Elean from West Greek. If *ē > ǟ would follow the compensatory lengthening, Elean would be a descendant of the doris severior group within West Greek, rather than a descendant from West Greek itself.

Bartonӗk defends this chronological order on the basis of evidence from the asymmetrical vowel system of Elean. The development of *ē > ǟ cannot, according to Bartonӗk, have been preceded by the first compensatory lengthening, since *ō would then be expected to develop - according to the rules of symmetry - into ᾱ. He, therefore, proposes *ē > ǟ to have preceded the compensatory lengthening, causing a pull chain, in which the ǟ (from *ē) pulls the result of the compensatory lengthening (ē₂) to its own former position in the vowel system.76

73 Cf. the examples given in chapter 1. Attic has βουλή < ō₂ and ὀκτώ < *ō, whereas in Elean, both these vowels are merged into one single ō: βōλά, ὀπτṓ. On the other hand, we find both ϝράτρα and ϝρɛ̄ ̄́τ[ρα] < *ē, whereas in words such as ἀτελɛ̄ ̄́ς, whose long ē is secondary, we only find the spelling <ɛ̄ > (and <η> after the introduction of the Ionic alphabet). This indicates that *ē developed into a phoneme different from ē₂.

74 Bartoněk 1972 : 157, 209-10.

75 Méndez Dosuna 1980 : 182, §I: Criterios de clasificación.

76 Bartoněk 1972 : 59.

Méndez Dosuna, however, points to the asymmetry of the vowel triangle, so that a development *ō > ᾱ parallel to that of *ē > ǟ is not necessarily to be expected. He in turn gives preference to a push chain, in which case the compensatory lengthening would have caused - and, therefore, preceded - the development *ē > ǟ in the following way: the appearance of a new set of long mid vowels forced the ancient *ē to be lowered.77 Hereby, Méndez Dosuna indicates that the change *ē > ǟ is not prior to the West Greek split into doris mitior and doris severior. It does, therefore, not put the split of Elean from West Greek back into the time prior to the compensatory lengthening processes.

Another argument for Bartonӗk to consider Elean an independent dialect within the West Greek group is easily refuted. According to him, the fact that Mycenaean uses different signs for denoting the syllables which contain a vocalism. This he relates to the Elean distinction between /ǟ/ and /ā/: both the Mycenaean and the Elean system could be ascribed to a pre-Doric substratum.78 However, it is beyond doubt that the Linear B signs a and a₂ represent different sound values, namely /a/ and /ha/ respectively, so that there is no similarity between these and the Elean distinction between /ǟ/ and /ā/. Apart from this, influence from the same substratum would be expected in the dialects from those areas where evidence of the use of Linear B has been found, which is, however, not the case.79

On the basis of the argument of the vocalic triange, Méndez Dosuna has sufficiently indicated that Elean did not separate itself from West Greek before the compensatory lengthening took place. Rather, it took the path of either doris severior or doris mitior. In the former case, its long vowel system would, after the development *ē > ǟ, end up with five vowels, which does not correspond to the data from the inscriptions. From this, we deduce that Elean took the path of doris mitior. Thus, it initially created a seven grade long vowel system, parallel to the other doris mitior dialects. Afterwards, but prior to the earliest inscriptions, it merged its two long ō vowels. Following either of these scenarios, it ended up with the six grade vowel system which is attested in the inscriptions.

2.5 The development *ӗ > α In the vowel system of Elean, there is a tendency of lowering not only long ē, but also its short pendant. As has become apparent from chapter 1, this is a general development in Northwest Greek when *ӗ is followed by ρ (cf. Elean φάρɛ̄ ν [25.5] ‘bear’, Locr. φάρειν, Delph. φάρεν).80 In Elean, however, the same change is found in other phonetic contexts. An example of this is the infinitive form γνōμαν [14.6], which we have seen in the discussion of this feature in paragraph 1.4.1, and further: μάν [22.3], σκευάōν [19.4], and ευσαβέοι [22.15].

This development is another reason for Bartonӗk to consider Elean a separate dialect branch within West Greek. According to him, this sound change is independent from its phonetic context. For that reason, he judges that it is probably related to the development *ē > ǟ and, therefore, another change that preceded the compensatory lengthening.81

77 Méndez Dosuna 1980: 187-8.

78 Bartoněk 1972 : 210.

79 Méndez Dosuna 1980 : 182-3.

80 Buck 1955 : 23.

81 Bartonӗk 1972 : 62.

Méndez Dosuna, however, states that the development *ӗ > α is, contrary to *ē > ǟ, conditioned by certain phonetic contexts, and that it is, for that reason, not related to the change *ē > ǟ in the long vowel system. Apart from those cases where α < *ӗ is preceded (or followed) by -r-, he tries to explain also the other instances in which this development is found.

First of all, in the forms γνōμαν and μάν, the α is followed by a nasal. According to Méndez Dosuna, *ӗ may here have been nasalised and, subsequently, opened to α.82 However, the α in μάν could also result from the shortening of *μᾱν, which contains a long vowel ᾱ. In this case, we would be dealing with the same doublet as Att. μήν and its shortened variant μέν.83 γνōμαν, secondly, could also be interpreted as an accusative singular form rather than as an infinitive, as Méndez Dosuna himself argued in 1984. However, in order to explain the supposed accusative γνōμαν syntactically, this requires that we have to assume that the author of the inscription omitted an infinitive εχɛ̄ ν.84 If these alternative interpretations of μαν and of γνōμαν are indeed correct, they would in both cases contain a long vowel ᾱ and, therefore, provide no indication of a development of short ε to ᾰ before a nasal.

A third instance of an alleged development of ε > α is seen in the gen. pl. σκευάōν [19.4]85 (= Att. σκευῶν ‘equipment’). In this case, Méndez Dosuna ascribes the α, where ε is expected, to the influence of preceding -w-: ε here developed into α due to a tendency of the two elements of a diphthong to differentiate maximally. A comparable development is seen in French, where -wę- developed into -wa- in the sixteenth and seventeenth century.86 However, this is not a valid parallel, since rising diphthongs were not present in Ancient Greek. Therefore, this comparison cannot be used as an argument in favour of a development -υε- > -υα- in Greek.

This leaves ευσαβέοι as the only form that speaks in favour of Bartonӗk’s view of *ӗ > α as a sound development independent of its context. Also in this form, we would expect ε instead of the actually attested α. Méndez Dosuna argues that in this form, the α has developed out of ε under the influence of a preceding -h-. However, in this inscription, α is preceded by σ rather than -h-. In order to explain this, Méndez Dosuna points to orthographic conventions. It is known that in Elean, intervocalic -σ- developed into -h-, after which it was either restored to -σ- or developed into -Ø-. In the aorist, however, -h- was kept rather than being restored to -σ-, because this tense was sufficiently characterised by other signs.87 This aspiration, however, ‘was felt a provincialism and ignored in the

82 Méndez Dosuna 1980: 191.

83 ‘[it] is functionally identical and related in meaning (…). μήν must have undergone vowel shortening as a result of a weakening of its function.’ (Beekes 2013 : 945).

84 Méndez Dosuna 1984 : 120.

85 Apart from σκευάōν, the form σκεύεα [19.12] is also attested, showing ε rather than α, which seems to contradict Méndez Dosuna’s proposed sound change. However, in this case ε is kept orthographically, Méndez Dosuna argues, in order to prevent a form **σκευαα.

86 Méndez Dosuna 1980 : 191-2.

87 Méndez Dosuna 1980 : 197.

spelling of the early and the majority of the later inscriptions.’88 Therefore, the underlying pronunciation of <σ> is [h], under the influence of which *ӗ developed into α.89

As an external evidence of an opening process of ε in contact with -h-, Méndez Dosuna points to the form αδεαλτώhαιε [30.12]. Also in this form, the α in the -hαιε ending is difficult to explain, since -σειε is the regular opt. aor. ending. The latter is also attested in Elean, in the form κατιαραύσειε [20.2].90 The traditional explanation of the form αδεαλτώhαιε is that it shows a blending of the aorist endings -σαι and -σειε. However, in case of an opening process of *ӗ after -h-, we can explain the α phonetically. This is preferable to assuming a process of analogy in order to give an explanation for one single form.

Therefore, Méndez Dosuna concludes, none of the abovementioned examples proves an independent development of ε > α. Rather, we are dealing with allophones dependent on certain phonetic contexts, i.e. -r-, -w-, -h- and, perhaps, -n-. For that reason, it cannot be related to the development *ē > ǟ.

As far as the contexts -r- and -n- are concerned, I am inclined to follow Méndez Dosuna’s opinion. The development of *ӗ to α in contact with ρ is a general one in the Northwest Greek dialects. And whereas the form μάν, since it may contain a long vowel, is no sufficient proof for the same development before a nasal, γνōμαν actually does indicate that we can assume such a change in Elean; for it is always preferable to explain an inscription on the basis of the actually attested material before assuming that ellipses have taken place.

However, I am less convinced of Méndez Dosuna’s proposition of a development *ӗ > α after -w- and -h-. As we have seen above, the origin of the α in σκευάōν is uncertain. The same goes, in my opinion, for the α in ευσαβέοι. My preference is to interpret the form the way it is written, before assuming a variation in spelling under the influence of sociolinguistic pressure. Therefore, the underlying sound value of <σ> is here, as it is anywhere else in the inscriptions, simply [s]. The forms σκευάōν and ευσαβέοι can, for that reason, only be explained by assuming a general development of *ӗ > α in Elean, which is possibly simultaneous with the lowering of *ē to ǟ.

In the next paragraph, our opinion to assume a development *ӗ > α under the influence of -r- and -n- will be strengthened by means of an analysis of another development in the short vowel system, i.e. that of *ĭ to ε.91

88 Buck 1955 : 55.

89 Méndez Dosuna 1980 : 199.

90 Méndez Dosuna 1980 : 189-90.

91 Méndez Dosuna 1980 : 197.

2.6 The development *i > ε As we have seen in paragraph 1.4.1, we find two instances of ε as the outcome of *ĭ in Elean: πόλερ ‘city’ [34.16] and 3rd sg. opt. βενέοι [4.1] ‘futuat’. As appears from the quotations, the phonetic contexts in which we find lowering of *ĭ to ε here are similar to the ones in which *ӗ is lowered to α: in the first form, *ĭ developed into ε before ρ (cf. μάν [22.3]); in the second, this development takes place before ν (cf. φάρɛ̄ ν [25.5]).

Τhe lowering in πόλερ is remarkable, since we are here dealing with a word-final ρ which originates from rhotacism (cf. paragraph 1.3.1). In the previous chapter, we have concluded that this rhotacism is a late feature, for only in the younger Elean inscriptions it is found consistently. Therefore, the development of *ĭ > ε must be a recent one. It cannot be related to the development *ӗ > α before ρ, since this is attested in Northwest Greek generally and must, therefore, be older.

According to Méndez Dosuna, we are indeed dealing with two phases of lowering. The former operates in archaic Elean and affects *ӗ. The latter operates only in late Elean and lowers *ĭ to ε. The development *ĭ > ε is, therefore, not one from which we may deduce dialectal relatedness.

3 Conclusion

In this chapter I will draw a conclusion concerning the genetic position of the Elean dialect within the Greek dialect system on the basis of the features that I chose to evaluate thoroughly in the previous chapter. Of course, compared to the overview of dialect features in chapter 1, this is a small selection on the basis of which to draw definite conclusions about the positioning of the dialect. Indeed, I do not deny that our insights might undergo change if all the Elean dialect features be considered and evaluated. Such an evaluation would, however, reach far beyond the scopes of this thesis.

We must furthermore keep in mind that for a genetic, diachronic classification, it is not the quantity of features shared with other dialects that is the determining factor. Rather, a single very old innovative change may tell us more about the genetic position of a dialect than a wide range of trivial features shared with neighboring dialects. Of course, this would only go for an innovation which cannot easily be ascribed to drift, i.e. a change that can manifest itself in different dialects independently. Indeed, we have concluded that this is the case with the consonant stem datives in -οις and -εσσι, and therefore rightly decided in paragraph 2.2 that this feature is not strong enough an indication of genetic relatedness.

On the other hand, we have seen that Elean has two morphological innovations which are actually reliable indications of relatedness between the dialects in which they are attested. The accusative plural ending -ες, which is found in consonant stem nouns, is one Elean shares with Locrian. With Locrian and Phocian it shares the middle participle which ends in /-ēmenos/. From this, it appears that these three dialects were still one group at the moment the new middle participle ending was created. After that, Phocian separated itself from this group before Elean and Locrian invented the accusative plural in -ες. This would explain the absence of this feature in Phocian.

This order of events is very well in accordance with the respective terminus ante quem we postulated for these developments. We have seen in the previous chapter that the middle participle in /-ēmenos/ is to be dated to 675 B.C. ultimately. The accusative plural in -ες, in its turn, could have been invented as late as the early 5th century B.C.

As far as the long vowel system is concerned, we assume, on the basis of the ancient innovations shared with Locrian and Phocian, that Elean took the path of doris mitior: it invented a system of seven long vowels as a result of the compensatory lengthening and contraction of like vowels, along with Phocian and Locrian. After that, and after it separated itself from the two last-mentioned dialects, it merged the two long ō vowels which its system contained, i.e. the originally long *ō, and the ō₂ which resulted from compensatory lengthening. Following these steps, it created the six grade long vowel system that is attested in the inscriptions.

For the developments in the short vowel system, we have to assume two phases. First, *ӗ is lowered before ρ at the moment the Northwest Greek dialects still form one group. Later, and only after the Northwest Greek dialects split up, the more specific, unconditioned development of *ӗ to α takes place in Elean. The lowering of *ĭ to ε, which is also specifically Elean, is to be ascribed only to this later phase as well.

In summary, the relative chronology of events is the following:

1. Innovation of a seven grade long vowel system in the doris mitior dialects, to which Northwest Greek belongs; 2. Lowering of *ӗ before ρ in Northwest Greek; 3. Creation of a middle participle in /-ēmenos/ in Phocian, Locrian, and Elean. Separation of these three from the other Northwest Greek dialects (i.e. Aetolian, Acarnanian, Epirotic); 4. Separation of Phocian from Locrian and Elean; 5. Innovation of an accusative plural ending -ες in consonant stem nouns in Locrian and Elean; 6. Merger of *ō and ō₂ and lowering of *ē to /ǟ/. Split of Elean from Locrian. Second lowering process (after rhotacism) of *ĭ and *ӗ in Elean.

In this sequence of events, the Elean dialect is not, as Bartonӗk argued, a separate descendant from West Greek. Rather, it initially took part in phonological and morphological developments which were not at all unique, before gradually inventing more specific features on its own. Via the path of doris mitior, it participated in the innovations which were common to all Northwest Greek dialects. Along with Phocian and Locrian, it created two morphological novelties before turning into a separate dialect. Only in this phase, it underwent its more specific changes such as rhotacism, which, according to Hesychius, caused its speakers to be regarded βαρβαρόφωνοι.92

92 Minon 2007 : 553.

4. Bibliography

Allen, W.S., Vox Graeca: a guide to the pronunciation of classical Greek, Cambridge: Cambrigde University Press 1968.

Bartoněk, A., Classification of the West Greek dialects at the time about 350 B.C., Amsterdam: Hakkert 1972.

Beekes, R.S.P., Etymological dictionary of Greek, Leiden: Brill 2010.

Buck, C.D., The Greek dialects, Chicago: The University of Chicago Press 1955.

Chantraine, P., Morphologie historique du grec, Paris: Klincksieck 1961.

García Ramón, J.L., “Anomale Schreibungen und Eleischer Vokalismus”, Akten des Internationalen Symposions „Elis und Achaia in der Antike“ (Athen, Mai 1989), 1992, 133-138.

García Ramón, J.L., “Griechisch ἱερός und seine Varianten, vedisch *iṣirá-”, in: Beekes, R.S.P., Lubotsky, A., Weitenberg, J.J.S., Rekonstruktion und Relative Chronologie, Akten der VIII. Fachtagung der Indogermanischen Gesellschaft, pp. 183-205, Innsbruck: Institut fü r Sprachwissenschaft, 1992.

García Ramón, J.L., “On the genetic classification of Ancient Greek Dialects: Comparative reconstruction versus Hypercriticism and Atomism at work”, in: Studies in Greek Linguistics 30 (2010) 219-236.

García Ramón, J.L., “Proportionale Analogie im Griechischen: Der Dativ pluralis der 3. Deklination in den äiolischen und westgriechischen Dialekten”, in: Meier-Brügger, M., Meiser G., Glotta: Zeitschrift fü r griechische und lateinische Sprache 68 (1990) 133-156, Göttingen: Vandenhoeck & Ruprecht.

Méndez Dosuna, J., “Clasificación dialectal y cronologia relativa: el dialecto eleo”, Studia Philologica Salmanticensia 4 (1980) 181-201.

Méndez Dosuna, J., Los dialectos dorios del noroeste – gramática y estudio dialectal, Ediciones Universidad de Salamanca, 1985.

Méndez Dosuna, J., “Northwest Greek”, in: Giannakis, G.K., Encyclopedia of Ancient Greek Language and Linguistics, Leiden: Brill 2013.

Méndez Dosuna, J., “Observaciones sobre la pretendida evolución ε > α y la flexión de los sustantivos –εύς en el dialecto eleo”, Emerita 52 (1984) 125-138.

Méndez Dosuna, J., “The Doric dialects”, in: Christidis, A.-F., A history of Ancient Greek from the beginnings to Late Antiquity, Cambridge: Cambridge University Press, 2007.

Miller, D. G., Ancient Greek Dialects and Early Authors, Introduction to the Dialect Mixture in Homer, with Notes on Lyric and Herodotus, Tübingen: De Gruyter, 2013.

Minon, S., “Le zêtacisme éléen. Z pour *d dans les inscription éléennes dialectales: trait phonétique ou graphique?”, Bulletin de la Société linguistique de Paris 93 (1998) 181-210.

Minon, S., Les inscriptions Éléennes dialectales (VIe – IIe siècle avant J.C.), Genève: Librairie Droz S.A., 2007.

Parker, H.M., “The linguistic case for the Aiolic migration reconsidered”, Hesperia 77 (2008) 431- 464.

Porzig, W., “Untersuchungen zu den altgriechischen Dialekten”, Indogermanische Forschungen 61 (1954) 147-169.

Risch, E., “Die Gliederung der griechischen Dialekte in neuer Sicht”, Museum Helveticum 12 (1955) 60-76.

Thevenot-Warelle, A., Le dialecte grec d’Elide, phoné tique et phonologie, Nancy: Presses Universitaires de Nancy, 1988.

Thumb, A., Kieckers, E., Handbuch der griechischen Dialekte I, Heidelberg: Carl Winters Universitätsbuchhandlung, 1932.